-------------------------------------------------------------------------------------------------------------------------------------------------
      name:  <unnamed>
       log:  C:\Users\alony\Dropbox\Alon\@ Research projects @\Peace process opinion\Perceptions of terrorism (with Avishay BSG)\Submissions\2022
> .09 PSRM (accepted)\2023.04 Final version\Data replication files\Replication log - Stata - Israel.log
  log type:  text
 opened on:  16 May 2023, 18:43:18

. do "C:\Users\alony\AppData\Local\Temp\STD7a38_000000.tmp"

. ***********************************************************************
. * Is Terrorism Necessarily Violent? Public Perceptions of Nonviolence * 
. * and Terrorism in Conflict Settings                                                              *
. *                                                                     *
. * Avishay Ben Sasson-Gordis & Alon Yakter                                                         *
. * Political Science Research and Methods (2023)                                           *
. *                                                                     *
. * Replication code - Israeli experiment                                                   * ***************************************************
> ********************
. * The analysis was performed with Stata 17.                                               *
. *                                                                                                                                         *
. * To load the data (the "use" command), users should place the .dta       *
. * file in their Stata working directory (or type the full file            *
. * paths as saved on their computer).                                              *
. ***********************************************************************
. 
. **********************************************
. * installing necessary user-written packages *
. **********************************************
. ssc install estout, replace
checking estout consistency and verifying not already installed...
all files already exist and are up to date.

. ssc install grstyle, replace
checking grstyle consistency and verifying not already installed...
all files already exist and are up to date.

. ssc install mplotoffset, replace
checking mplotoffset consistency and verifying not already installed...
all files already exist and are up to date.

. net install grc1leg2.pkg, from (http://digital.cgdev.org/doc/stata/MO/Misc/) replace
checking grc1leg2 consistency and verifying not already installed...
all files already exist and are up to date.

. 
. *****************************
. * Load Israeli data             *
. *****************************
. use "BSG & Yakter PSRM - Israel survey-experiment data.dta", replace

. 
. *******************************
. * Descriptive plot (Figure 1) *
. *******************************
. grstyle init

. grstyle set legend, nobox

. grstyle set plain, nogrid box

. graph bar (percent), ///
>         over(is_terror, gap(*0.5) nolab) ///
>         bar(1 ,color(black)) ytitle("Percent") ylabel(,nogrid) ///
>         text(50 1 "Mean=7.77", place(e) size(medlarge)) text(46 1 "S.D.=2.73", place(e) size(medlarge)) ///
>         title("Denotation") name(is_terror_bar, replace)

. graph bar (percent), ///
>         over(is_illegitimate, gap(*0.5) nolab) ///
>         bar(1 ,color(black)) ytitle("Percent") ylabel(,nogrid) ///
>         text(50 1 "Mean=8.06", place(e) size(medlarge)) text(46 1 "S.D.=2.6", place(e) size(medlarge)) ///
>         title("Illegitimacy") name(is_illegitimate_bar, replace)

. graph bar (percent), ///
>         over(use_force, gap(*0.5) nolab) ///
>         bar(1 ,color(black)) ytitle("Percent") ylabel(,nogrid) ///
>         text(50 1 "Mean=7.92", place(e) size(medlarge)) text(46 1 "S.D.=2.66", place(e) size(medlarge)) ///
>         title("Use of Force") name(use_force_bar, replace)

. graph combine is_terror_bar is_illegitimate_bar use_force_bar, ///
>         col(3) ycommon  xsize(3) ysize(1) scale(1.8)

. graph export Figure1.pdf
file Figure1.pdf saved as PDF format

. 
. **************************************
. * Influence of action type (Table 2) *
. **************************************
. eststo clear

. qui eststo: reg is_terror i.tr_nonviolence attention

. qui eststo: reg is_terror i.tr_action attention

. qui eststo: reg is_illegitimate i.tr_nonviolence attention

. qui eststo: reg is_illegitimate i.tr_action attention

. qui eststo: reg use_force i.tr_nonviolence attention

. qui eststo: reg use_force i.tr_action attention

. esttab using Table2.rtf, ///
>         b(3) se(3) r2(3) nogaps compress ///
>         drop(1.tr_action 0*) ///
>         star(+ 0.1 * 0.05 ** 0.01 *** 0.001) ///
>         order(_cons 1.tr_nonviolence 2.tr_action 3.tr_action 4.tr_action attention) ///
>         coeflabels(1.tr_nonviolence Nonviolence ///
>                 2.tr_action Economic_action 3.tr_action Legal_action 4.tr_action Construct_action /// 
>                 attention Attention _cons Constant-Violence) ///
>         mgroups("Terror Denotation" "Illegitimacy" "Use Force", pattern(1 0 1 0 1 0)) ///
>         mtitles("Violence" "Action" "Violence" "Action" "Violence" "Action") 
(output written to Table2.rtf)

. * wald tests for action type by harm
. qui reg is_terror i.tr_action attention

. test i2.tr_action = i3.tr_action

 ( 1)  2.tr_action - 3.tr_action = 0

       F(  1,  2000) =    4.51
            Prob > F =    0.0338

. test i2.tr_action = i4.tr_action

 ( 1)  2.tr_action - 4.tr_action = 0

       F(  1,  2000) =   22.15
            Prob > F =    0.0000

. test i3.tr_action = i4.tr_action

 ( 1)  3.tr_action - 4.tr_action = 0

       F(  1,  2000) =    6.51
            Prob > F =    0.0108

. qui reg is_illegitimate i.tr_action attention

. test i2.tr_action = i3.tr_action

 ( 1)  2.tr_action - 3.tr_action = 0

       F(  1,  2000) =    5.83
            Prob > F =    0.0158

. test i2.tr_action = i4.tr_action

 ( 1)  2.tr_action - 4.tr_action = 0

       F(  1,  2000) =    1.57
            Prob > F =    0.2110

. test i3.tr_action = i4.tr_action

 ( 1)  3.tr_action - 4.tr_action = 0

       F(  1,  2000) =    1.36
            Prob > F =    0.2441

. qui reg use_force i.tr_action attention

. test i2.tr_action = i3.tr_action

 ( 1)  2.tr_action - 3.tr_action = 0

       F(  1,  2000) =    3.08
            Prob > F =    0.0795

. test i2.tr_action = i4.tr_action

 ( 1)  2.tr_action - 4.tr_action = 0

       F(  1,  2000) =    0.14
            Prob > F =    0.7073

. test i3.tr_action = i4.tr_action

 ( 1)  3.tr_action - 4.tr_action = 0

       F(  1,  2000) =    4.50
            Prob > F =    0.0340

. 
. *****************************************************
. * Influence of action type - by ideology (Figure 2) *
. *****************************************************
. grstyle init

. grstyle set legend, nobox

. grstyle set plain, nogrid box

. qui reg is_terror i.tr_nonviolence##i.vote_bloc_rcl i.attention if vote_bloc_rcl!=4

. margins, at(vote_bloc_rcl=(1(1)3) tr_nonviolence=(0 1))

Predictive margins                                       Number of obs = 1,583
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_nonviolence = 0
       vote_bloc_rcl  = 1
2._at: tr_nonviolence = 0
       vote_bloc_rcl  = 2
3._at: tr_nonviolence = 0
       vote_bloc_rcl  = 3
4._at: tr_nonviolence = 1
       vote_bloc_rcl  = 1
5._at: tr_nonviolence = 1
       vote_bloc_rcl  = 2
6._at: tr_nonviolence = 1
       vote_bloc_rcl  = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |   8.860558   .3230847    27.42   0.000     8.226837    9.494279
          2  |   9.068723   .2485128    36.49   0.000     8.581272    9.556173
          3  |   9.232857   .1431185    64.51   0.000     8.952134     9.51358
          4  |   4.307671   .1955494    22.03   0.000     3.924107    4.691235
          5  |   6.471896   .1431108    45.22   0.000     6.191189    6.752604
          6  |    8.21758   .0839101    97.93   0.000     8.052993    8.382168
------------------------------------------------------------------------------

. mplotoffset, recast(scatter) ///
>         plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>         plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>         legend(order(3 "Violence" 4 "Non-violence" )) ///
>         yscale(range(2 10)) ylabel(2(2)10, labsize(medsmall)) ///
>         xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(5.5, lpattern(dash) lcolor(gr6)) ///
>         title("Denotation", size(large)) ///
>         name(terror_violence_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_nonviolence
(note:  named style gr6 not found in class color, default attributes used)

. qui reg is_illegitimate i.tr_nonviolence##i.vote_bloc_rcl i.attention if vote_bloc_rcl!=4

. margins, at(vote_bloc_rcl=(1(1)3) tr_nonviolence=(0 1))

Predictive margins                                       Number of obs = 1,583
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_nonviolence = 0
       vote_bloc_rcl  = 1
2._at: tr_nonviolence = 0
       vote_bloc_rcl  = 2
3._at: tr_nonviolence = 0
       vote_bloc_rcl  = 3
4._at: tr_nonviolence = 1
       vote_bloc_rcl  = 1
5._at: tr_nonviolence = 1
       vote_bloc_rcl  = 2
6._at: tr_nonviolence = 1
       vote_bloc_rcl  = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |   8.703904   .3297108    26.40   0.000     8.057186    9.350622
          2  |   8.859353   .2536094    34.93   0.000     8.361906    9.356801
          3  |   9.205186   .1460537    63.03   0.000     8.918706    9.491666
          4  |   5.451688   .1995599    27.32   0.000     5.060257    5.843119
          5  |   7.163159   .1460458    49.05   0.000     6.876695    7.449624
          6  |   8.369679   .0856309    97.74   0.000     8.201717    8.537642
------------------------------------------------------------------------------

. mplotoffset, recast(scatter) ///
>         plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>         plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>         legend(order(3 "Violence" 4 "Non-violence" )) ///
>         yscale(range(2 10)) ylabel(2(2)10, labsize(medsmall)) ///
>         xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(5.5, lpattern(dash) lcolor(gr6)) ///
>         title("Illegitimacy", size(large)) ///
>         name(legitimate_violence_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_nonviolence
(note:  named style gr6 not found in class color, default attributes used)

. qui reg use_force i.tr_nonviolence##i.vote_bloc_rcl i.attention if vote_bloc_rcl!=4

. margins, at(vote_bloc_rcl=(1(1)3) tr_nonviolence=(0 1))

Predictive margins                                       Number of obs = 1,583
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_nonviolence = 0
       vote_bloc_rcl  = 1
2._at: tr_nonviolence = 0
       vote_bloc_rcl  = 2
3._at: tr_nonviolence = 0
       vote_bloc_rcl  = 3
4._at: tr_nonviolence = 1
       vote_bloc_rcl  = 1
5._at: tr_nonviolence = 1
       vote_bloc_rcl  = 2
6._at: tr_nonviolence = 1
       vote_bloc_rcl  = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |   8.552792   .3155101    27.11   0.000     7.933929    9.171656
          2  |   8.750711   .2426864    36.06   0.000     8.274689    9.226733
          3  |   9.447918   .1397631    67.60   0.000     9.173777     9.72206
          4  |   4.540148   .1909648    23.77   0.000     4.165576     4.91472
          5  |   6.732292   .1397556    48.17   0.000     6.458166    7.006418
          6  |   8.321317   .0819428   101.55   0.000     8.160589    8.482045
------------------------------------------------------------------------------

. mplotoffset, recast(scatter) ///
>         plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>         plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>         legend(order(3 "Violence" 4 "Non-violence" )) ///
>         yscale(range(2 10)) ylabel(2(2)10, labsize(medsmall)) ///
>         xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(5.5, lpattern(dash) lcolor(gr6)) ///
>         title("Use of Force", size(large)) ///
>         name(force_violence_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_nonviolence
(note:  named style gr6 not found in class color, default attributes used)

. grc1leg2 terror_violence_pv legitimate_violence_pv force_violence_pv ///
>         , rows(1) xcommon ycommon labsize(small) ///
>         b2title("Vote by Partisan Bloc", size(medsmall)) ///
>         l1title("Predicted Values",  size(medsmall)) ring(2) ///
>         xsize(6) ysize(3) scale(1.4) ///
>         name(combine_violence, replace)
-grc1leg2- working...
(note:  named style gr6 not found in class color, default attributes used)
(note:  named style gr6 not found in class color, default attributes used)
(note:  named style gr6 not found in class color, default attributes used)
Warning: Because the legend borrowed for the combined graph is not "complete",
automatic resizing of legend markers is disabled and grc1leg2 ignores the option -lmsize-.
See the discussion of "known issues" in grc1leg2's help file here.

. graph export Figure2.pdf
file Figure2.pdf saved as PDF format

. 
. ***********************************
. * Influence of labeling (Table 3) *
. ***********************************
. eststo clear

. qui eststo: reg is_terror i.tr_frame attention if tr_nonviolence==1

. qui eststo: reg is_illegitimate i.tr_frame attention if tr_nonviolence==1

. qui eststo: reg use_force i.tr_frame attention if tr_nonviolence==1

. esttab using Table3.rtf, ///
>         b(3) se(3) r2(3) nogaps compress ///
>         drop(0*) ///
>         star(+ 0.1 * 0.05 ** 0.01 *** 0.001) ///
>         order(_cons 1.tr_frame attention) ///
>         coeflabels(1.tr_frame Terror-Label attention Attention _cons Constant-No_Label) ///
>         mtitles("Terror Denotation" "Legitimacy" "Use Force") 
(output written to Table3.rtf)

.                 
. **************************************************
. * Influence of labeling - by ideology (Figure 3) *
. **************************************************
. qui reg is_terror i.tr_frame##i.vote_bloc_rcl i.attention if vote_bloc_rcl!=4 & tr_nonviolence==1

. margins, at(vote_bloc_rcl=(1(1)3) tr_frame=(0 1))

Predictive margins                                       Number of obs = 1,178
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_frame      = 0
       vote_bloc_rcl = 1
2._at: tr_frame      = 0
       vote_bloc_rcl = 2
3._at: tr_frame      = 0
       vote_bloc_rcl = 3
4._at: tr_frame      = 1
       vote_bloc_rcl = 1
5._at: tr_frame      = 1
       vote_bloc_rcl = 2
6._at: tr_frame      = 1
       vote_bloc_rcl = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |     4.1959   .2926516    14.34   0.000      3.62172     4.77008
          2  |    6.10201   .2074172    29.42   0.000     5.695059    6.508961
          3  |   8.077461   .1287741    62.73   0.000     7.824807    8.330115
          4  |   4.434316   .2964663    14.96   0.000     3.852652    5.015981
          5  |    6.90775   .2245567    30.76   0.000     6.467171    7.348328
          6  |   8.343555   .1240289    67.27   0.000     8.100212    8.586899
------------------------------------------------------------------------------

. mplotoffset, recast(scatter) ///
>         plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>         plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>         legend(order(3 "No Label" 4 "Terror Label" )) ///
>         yscale(range(5 10)) ylabel(5(2)10, labsize(medsmall)) ///
>         xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(5.5, lpattern(dash) lcolor(gr6)) ///
>         title("Denotation", size(large)) ///
>         name(terror_label_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_frame
(note:  named style gr6 not found in class color, default attributes used)

. qui reg is_illegitimate i.tr_frame##i.vote_bloc_rcl i.attention if vote_bloc_rcl!=4 & tr_nonviolence==1

. margins, at(vote_bloc_rcl=(1(1)3) tr_frame=(0 1))

Predictive margins                                       Number of obs = 1,178
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_frame      = 0
       vote_bloc_rcl = 1
2._at: tr_frame      = 0
       vote_bloc_rcl = 2
3._at: tr_frame      = 0
       vote_bloc_rcl = 3
4._at: tr_frame      = 1
       vote_bloc_rcl = 1
5._at: tr_frame      = 1
       vote_bloc_rcl = 2
6._at: tr_frame      = 1
       vote_bloc_rcl = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |   5.398937   .2959581    18.24   0.000      4.81827    5.979604
          2  |   7.073458   .2097607    33.72   0.000     6.661909    7.485006
          3  |   8.339317   .1302291    64.04   0.000     8.083808    8.594825
          4  |   5.506901   .2998159    18.37   0.000     4.918664    6.095137
          5  |   7.267986   .2270938    32.00   0.000      6.82243    7.713542
          6  |   8.396771   .1254302    66.94   0.000     8.150678    8.642864
------------------------------------------------------------------------------

. mplotoffset, recast(scatter) ///
>         plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>         plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>         legend(order(3 "No Label" 4 "Terror Label" )) ///
>         yscale(range(5 10)) ylabel(5(2)10, labsize(medsmall)) ///
>         xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(5.5, lpattern(dash) lcolor(gr6)) ///
>         title("Illegitimacy", size(large)) ///
>         name(legitimate_label_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_frame
(note:  named style gr6 not found in class color, default attributes used)

. qui reg use_force i.tr_frame##i.vote_bloc_rcl i.attention if vote_bloc_rcl!=4 & tr_nonviolence==1

. margins, at(vote_bloc_rcl=(1(1)3) tr_frame=(0 1))

Predictive margins                                       Number of obs = 1,178
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_frame      = 0
       vote_bloc_rcl = 1
2._at: tr_frame      = 0
       vote_bloc_rcl = 2
3._at: tr_frame      = 0
       vote_bloc_rcl = 3
4._at: tr_frame      = 1
       vote_bloc_rcl = 1
5._at: tr_frame      = 1
       vote_bloc_rcl = 2
6._at: tr_frame      = 1
       vote_bloc_rcl = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |   4.570284   .2873229    15.91   0.000     4.006559    5.134009
          2  |   6.318726   .2036404    31.03   0.000     5.919185    6.718267
          3  |   8.306618   .1264294    65.70   0.000     8.058565    8.554671
          4  |   4.520165   .2910681    15.53   0.000     3.949091    5.091238
          5  |   7.219176   .2204679    32.74   0.000      6.78662    7.651733
          6  |   8.331138   .1217705    68.42   0.000     8.092226    8.570051
------------------------------------------------------------------------------

. mplotoffset, recast(scatter) ///
>         plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>         plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>         legend(order(3 "No Label" 4 "Terror Label" )) ///
>         yscale(range(5 10)) ylabel(5(2)10, labsize(medsmall)) ///
>         xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(5.5, lpattern(dash) lcolor(gr6)) ///
>         title("Use of Force", size(large)) ///
>         name(force_label_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_frame
(note:  named style gr6 not found in class color, default attributes used)

. grc1leg2 terror_label_pv legitimate_label_pv force_label_pv ///
>         , rows(1) xcommon ycommon labsize(small) ///
>         b2title("Vote by Partisan Bloc", size(medsmall)) ///
>         l1title("Predicted Values",  size(medsmall)) ring(2) ///
>         xsize(6) ysize(3) scale(1.4) ///
>         name(combine_label, replace)
-grc1leg2- working...
(note:  named style gr6 not found in class color, default attributes used)
(note:  named style gr6 not found in class color, default attributes used)
(note:  named style gr6 not found in class color, default attributes used)
Warning: Because the legend borrowed for the combined graph is not "complete",
automatic resizing of legend markers is disabled and grc1leg2 ignores the option -lmsize-.
See the discussion of "known issues" in grc1leg2's help file here.

. graph export Figure3.pdf
file Figure3.pdf saved as PDF format

. 
. *******************************************
. * Appendix: Sample tests and descriptives *
. *******************************************
. * balance tests by demography (Table A2)
. * note: Stata orders the table columns differently, we have reordered them manually in the appendix.
. ****************************************
. eststo clear

. qui eststo: mlogit treatment_group i.sex i.region i.agegr i.edu i.relid i.vote_bloc_rcl i.attention, b(2)

. esttab using TableA2.rtf, ///
>         b(3) unstack nogaps noomitted not label noobs drop(1* 0* _cons)
(output written to TableA2.rtf)

. 
. * power analysis (scores used for Table A3)
. *******************************************
. retrodesign 0.3 0.5 1 1.5, se(0.15) alpha(0.05) df(245)

      Effect |     Power    S-error    M-error 
-------------+---------------------------------
         0.3 |    0.5121     0.0001     1.3925 
         0.5 |    0.9130     0.0000     1.0480 
           1 |    1.0000     0.0000     1.0023 
         1.5 |    1.0000     0.0000     1.0016 
-----------------------------------------------

Note:
- Power is the probability that the statistical test correctly rejects the null hypothesis.
- Type-S (sign) error is the probability of the sign being in the opposite direction of the effect size.
- Type-M (magnitude) error is the factor by which the magnitude of the effect size might be exaggerated.

. 
. * descriptive stats 
. *******************
. * summary stats (scores used for Table C1)
. summarize is_terror is_illegitimate use_force attention i.vote_bloc_rcl if is_terror!=.

    Variable |        Obs        Mean    Std. dev.       Min        Max
-------------+---------------------------------------------------------
   is_terror |      2,005    7.766085    2.732894          1         10
is_illegit~e |      2,005    8.060848      2.6036          1         10
   use_force |      2,005      7.9202    2.657159          1         10
   attention |      2,005    .9376559    .2418399          0          1
             |
vote_bloc_~l |
       left  |      2,005    .0967581    .2957019          0          1
-------------+---------------------------------------------------------
     center  |      2,005    .1760599    .3809661          0          1
      right  |      2,005    .5167082    .4998454          0          1
did not v..  |      2,005    .2104738    .4077469          0          1

. * distribution graphs (Figure C1)
. graph bar (percent), ///
>         over(is_terror, gap(*0.5) nolab) ///
>         bar(1 ,color(black)) ytitle("Percent") ylabel(,nogrid) ///
>         title("Denotation") name(is_terror_bar, replace)

. graph bar (percent), ///
>         over(is_illegitimate, gap(*0.5) nolab) ///
>         bar(1 ,color(black)) ytitle("Percent") ylabel(,nogrid) ///
>         title("Illegitimacy") name(is_illegitimate_bar, replace)

. graph bar (percent), ///
>         over(use_force, gap(*0.5) nolab) ///
>         bar(1 ,color(black)) ytitle("Percent") ylabel(,nogrid) ///
>         title("Use of Force") name(use_force_bar, replace)

. graph bar (percent), ///
>         over(vote_bloc_rcl, gap(*0.5) relabel(1 "Left" 2 "Center" 3 "Right" 4 `""Not" "Voted""')) ///
>         bar(1 ,color(black)) ytitle("Percent") ylabel(,nogrid) ///
>         title("Vote by Partisan Bloc") name(vote_bloc_rcl_bar, replace)

. graph bar (percent), ///
>         over(attention, gap(*0.5) relabel(1 `""Failed" "Check""' 2 `""Passed" "Check""')) ///
>         bar(1 ,color(black)) ytitle("Percent") ylabel(,nogrid) ///
>         title("Attention Check") name(attention_bar, replace)

. graph combine is_terror_bar is_illegitimate_bar use_force_bar vote_bloc_rcl_bar attention_bar, ///
>         col(3) ycommon

. graph export FigureC1.pdf
file FigureC1.pdf saved as PDF format

. 
. ***************************************
. * Appendix: full models behind graphs *
. ***************************************
. * violence vs nonviolence interacted with partisanship (Table D1)
. eststo clear

. qui eststo: reg is_terror i.tr_nonviolence##i.vote_bloc_rcl attention if vote_bloc_rcl!=4

. qui eststo: reg is_illegitimate i.tr_nonviolence##i.vote_bloc_rcl attention if vote_bloc_rcl!=4

. qui eststo: reg use_force i.tr_nonviolence##i.vote_bloc_rcl attention if vote_bloc_rcl!=4

. esttab using TableD1.rtf, ///
>         b(3) se(3) r2(3) nogaps compress ///
>         drop(1.vote_bloc_rcl 1.tr_nonviolence#1.vote_bloc_rcl 0*) ///
>         star(+ 0.1 * 0.05 ** 0.01 *** 0.001) ///
>         order(_cons 1.tr_nonviolence 2.vote_bloc_rcl 3.vote_bloc_rcl 1.tr_nonviolence#2.vote_bloc_rcl ///
>                 1.tr_nonviolence#3.vote_bloc_rcl attention) ///
>         coeflabels(1.tr_nonviolence Nonviolence 2.vote_bloc_rcl Center 3.vote_bloc_rcl Right ///
>                 1.tr_nonviolence#2.vote_bloc_rcl NonviolenceXCenter 1.tr_nonviolence#3.vote_bloc_rcl ///
>                 NonviolenceXRight attention Attention _cons Constant-Violence-Left) ///
>         mtitles("Terror Denotation" "Illegitimacy" "Use Force") 
(output written to TableD1.rtf)

.         
. * terror label interacted with partisanship (Table D2)
. eststo clear

. qui eststo: reg is_terror i.tr_frame##i.vote_bloc_rcl attention if vote_bloc_rcl!=4 & tr_nonviolence==1

. qui eststo: reg is_illegitimate i.tr_frame##i.vote_bloc_rcl attention if vote_bloc_rcl!=4 & tr_nonviolence==1

. qui eststo: reg use_force i.tr_frame##i.vote_bloc_rcl attention if vote_bloc_rcl!=4 & tr_nonviolence==1

. esttab using TableD2.rtf, ///
>         b(3) se(3) r2(3) nogaps compress ///
>         drop(1.vote_bloc_rcl 1.tr_frame#1.vote_bloc_rcl 0*) ///
>         star(+ 0.1 * 0.05 ** 0.01 *** 0.001) ///
>         order(_cons 1.tr_frame 2.vote_bloc_rcl 3.vote_bloc_rcl 1.tr_frame#2.vote_bloc_rcl ///
>                 1.tr_frame#3.vote_bloc_rcl attention) ///
>         coeflabels(1.tr_frame Terror_Label 2.vote_bloc_rcl Center 3.vote_bloc_rcl Right ///
>                 1.tr_frame#2.vote_bloc_rcl Terror_LabelXCenter 1.tr_frame#3.vote_bloc_rcl ///
>                 Terror_LabelXRight attention Attention _cons Constant-No_Label-Left) ///
>         mtitles("Terror Denotation" "Illegitimacy" "Use Force")
(output written to TableD2.rtf)

. 
. * logit analysis for substantive implications of framing effect (Table D3)
. eststo clear

. qui eststo: logit force_hi_dummy i.tr_frame##i.vote_bloc_rcl attention if vote_bloc_rcl!=4 & tr_nonviolence==1

. esttab using TableD3.rtf, ///
>         b(3) se(3) scalars(r2_p) nogaps compress nomtitles nodepvars ///
>         drop(1.vote_bloc_rcl 1.tr_frame#1.vote_bloc_rcl 0*) ///
>         star(+ 0.1 * 0.05 ** 0.01 *** 0.001) ///
>         order(_cons 1.tr_frame 2.vote_bloc_rcl 3.vote_bloc_rcl 1.tr_frame#2.vote_bloc_rcl ///
>                 1.tr_frame#3.vote_bloc_rcl attention) ///
>         coeflabels(1.tr_frame Terror_Label 2.vote_bloc_rcl Center 3.vote_bloc_rcl Right ///
>                 1.tr_frame#2.vote_bloc_rcl Terror_LabelXCenter 1.tr_frame#3.vote_bloc_rcl ///
>                 Terror_LabelXRight attention Attention _cons Constant-No_Label-Left) 
(output written to TableD3.rtf)

. * post-estimaiton predicted probabilities (scores used for Table D4)
. margins, at(vote_bloc_rcl=(1(1)3) tr_frame=(0 1))

Predictive margins                                       Number of obs = 1,178
Model VCE: OIM

Expression: Pr(force_hi_dummy), predict()
1._at: tr_frame      = 0
       vote_bloc_rcl = 1
2._at: tr_frame      = 0
       vote_bloc_rcl = 2
3._at: tr_frame      = 0
       vote_bloc_rcl = 3
4._at: tr_frame      = 1
       vote_bloc_rcl = 1
5._at: tr_frame      = 1
       vote_bloc_rcl = 2
6._at: tr_frame      = 1
       vote_bloc_rcl = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |   .1627399   .0429876     3.79   0.000     .0784856    .2469941
          2  |   .3837654   .0405065     9.47   0.000     .3043741    .4631566
          3  |    .736462   .0227618    32.36   0.000     .6918498    .7810743
          4  |    .201032   .0479224     4.19   0.000     .1071059    .2949581
          5  |   .5320746   .0449797    11.83   0.000     .4439161    .6202331
          6  |   .7286095   .0221206    32.94   0.000     .6852539    .7719651
------------------------------------------------------------------------------

. 
. ******************************
. * Appendix: robustness tests *
. ******************************
. * adding demographic controls
. *****************************
. * action and harm - regression (Table E1)
. eststo clear

. qui eststo: reg is_terror i.tr_nonviolence attention sex agegr i.relid edu oleh i.region  

. qui eststo: reg is_terror i.tr_action attention sex agegr i.relid edu oleh i.region 

. qui eststo: reg is_illegitimate i.tr_nonviolence attention sex agegr i.relid edu oleh i.region 

. qui eststo: reg is_illegitimate i.tr_action attention sex agegr i.relid edu oleh i.region 

. qui eststo: reg use_force i.tr_nonviolence attention sex agegr i.relid edu oleh i.region 

. qui eststo: reg use_force i.tr_action attention sex agegr i.relid edu oleh i.region 

. esttab using TableE1.rtf, ///
>         b(3) se(3) r2(3) nogaps compress ///
>         drop(1.tr_action 1.relid 1.region 0*) ///
>         star(+ 0.1 * 0.05 ** 0.01 *** 0.001) ///
>         order(_cons 1.tr_nonviolence 2.tr_action 3.tr_action 4.tr_action attention) ///
>         coeflabels(1.tr_nonviolence Nonviolence ///
>                 2.tr_action Economic_action 3.tr_action Legal_action 4.tr_action construct_action /// 
>                 1.attention Attention _cons Constant-Violence sex Gender-Female agegr Age-Group ///
>                 2.relid Religious-Id-Traditional 3.relid Religious-Id-Religious 4.relid ///
>                 Religious-Id-Haredi edu Education oleh Immigrant 2.region Region-North 3.region Region-Haifa ///
>                 4.region Region-Center 5.region Region-Tel-Aviv 6.region Region-South 7.region Region-West-Bank) ///
>         mgroups("Terror Denotation" "Illegitimacy" "Use Force", pattern(1 0 1 0 1 0)) ///
>         mtitles("Violence" "Action" "Violence" "Action" "Violence" "Action") 
(output written to TableE1.rtf)

. * action and harm - hetereognous effects, predicted value graphs (Figure E1)
.         qui reg is_terror i.tr_nonviolence##i.vote_bloc_rcl i.attention sex agegr i.relid edu oleh i.region if vote_bloc_rcl!=4

.         margins, at(vote_bloc_rcl=(1(1)3) tr_nonviolence=(0 1))

Predictive margins                                       Number of obs = 1,583
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_nonviolence = 0
       vote_bloc_rcl  = 1
2._at: tr_nonviolence = 0
       vote_bloc_rcl  = 2
3._at: tr_nonviolence = 0
       vote_bloc_rcl  = 3
4._at: tr_nonviolence = 1
       vote_bloc_rcl  = 1
5._at: tr_nonviolence = 1
       vote_bloc_rcl  = 2
6._at: tr_nonviolence = 1
       vote_bloc_rcl  = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |   9.078437   .3264658    27.81   0.000      8.43808    9.718794
          2  |   9.219094   .2533609    36.39   0.000     8.722131    9.716057
          3  |   9.176894   .1432664    64.05   0.000      8.89588    9.457909
          4  |   4.544142   .2056165    22.10   0.000     4.140828    4.947455
          5  |     6.6309   .1502325    44.14   0.000     6.336222    6.925578
          6  |   8.106754   .0894489    90.63   0.000     7.931302    8.282207
------------------------------------------------------------------------------

.         mplotoffset, recast(scatter) ///
>                 plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>                 plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>                 legend(order(3 "Violence" 4 "Non-violence" )) ///
>                 yscale(range(2 10)) ylabel(2(2)10, labsize(medsmall)) ///
>                 xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>                 ytitle("")      xtitle("")  ///
>                 yline(5.5, lpattern(dash) lcolor(gr6)) ///
>                 title("Denotation", size(large)) ///
>                 name(terror_violence_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_nonviolence
(note:  named style gr6 not found in class color, default attributes used)

.         qui reg is_illegitimate i.tr_nonviolence##i.vote_bloc_rcl i.attention sex agegr i.relid edu oleh i.region if vote_bloc_rcl!=4

.         margins, at(vote_bloc_rcl=(1(1)3) tr_nonviolence=(0 1))

Predictive margins                                       Number of obs = 1,583
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_nonviolence = 0
       vote_bloc_rcl  = 1
2._at: tr_nonviolence = 0
       vote_bloc_rcl  = 2
3._at: tr_nonviolence = 0
       vote_bloc_rcl  = 3
4._at: tr_nonviolence = 1
       vote_bloc_rcl  = 1
5._at: tr_nonviolence = 1
       vote_bloc_rcl  = 2
6._at: tr_nonviolence = 1
       vote_bloc_rcl  = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |   8.931083     .33318    26.81   0.000     8.277556     9.58461
          2  |   9.014023   .2585716    34.86   0.000     8.506839    9.521206
          3  |   9.155582   .1462129    62.62   0.000     8.868788    9.442376
          4  |   5.613153   .2098453    26.75   0.000     5.201545    6.024761
          5  |   7.345528   .1533222    47.91   0.000     7.044789    7.646266
          6  |   8.261333   .0912886    90.50   0.000     8.082272    8.440394
------------------------------------------------------------------------------

.         mplotoffset, recast(scatter) ///
>                 plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>                 plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>                 legend(order(3 "Violence" 4 "Non-violence" )) ///
>                 yscale(range(2 10)) ylabel(2(2)10, labsize(medsmall)) ///
>                 xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>                 ytitle("")      xtitle("")  ///
>                 yline(5.5, lpattern(dash) lcolor(gr6)) ///
>                 title("Illegitimacy", size(large)) ///
>                 name(legitimate_violence_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_nonviolence
(note:  named style gr6 not found in class color, default attributes used)

.         qui reg use_force i.tr_nonviolence##i.vote_bloc_rcl i.attention sex agegr i.relid edu oleh i.region if vote_bloc_rcl!=4

.         margins, at(vote_bloc_rcl=(1(1)3) tr_nonviolence=(0 1))

Predictive margins                                       Number of obs = 1,583
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_nonviolence = 0
       vote_bloc_rcl  = 1
2._at: tr_nonviolence = 0
       vote_bloc_rcl  = 2
3._at: tr_nonviolence = 0
       vote_bloc_rcl  = 3
4._at: tr_nonviolence = 1
       vote_bloc_rcl  = 1
5._at: tr_nonviolence = 1
       vote_bloc_rcl  = 2
6._at: tr_nonviolence = 1
       vote_bloc_rcl  = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |   8.754931   .3183813    27.50   0.000     8.130431     9.37943
          2  |   8.929507   .2470867    36.14   0.000      8.44485    9.414163
          3  |   9.389683   .1397186    67.20   0.000     9.115627    9.663739
          4  |   4.785472   .2005246    23.86   0.000     4.392146    5.178798
          5  |   6.911246   .1465121    47.17   0.000     6.623865    7.198627
          6  |   8.200601   .0872338    94.01   0.000     8.029494    8.371709
------------------------------------------------------------------------------

.         mplotoffset, recast(scatter) ///
>                 plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>                 plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>                 legend(order(3 "Violence" 4 "Non-violence" )) ///
>                 yscale(range(2 10)) ylabel(2(2)10, labsize(medsmall)) ///
>                 xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>                 ytitle("")      xtitle("")  ///
>                 yline(5.5, lpattern(dash) lcolor(gr6)) ///
>                 title("Use of Force", size(large)) ///
>                 name(force_violence_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_nonviolence
(note:  named style gr6 not found in class color, default attributes used)

.         grc1leg2 terror_violence_pv legitimate_violence_pv force_violence_pv ///
>                 , rows(1) xcommon ycommon labsize(small) ///
>                 b2title("Vote by Partisan Bloc", size(medsmall)) ///
>                 l1title("Predicted Values",  size(medsmall)) ring(2) ///
>                 xsize(6) ysize(3) scale(1.4) ///
>                 name(combine_violence, replace)
-grc1leg2- working...
(note:  named style gr6 not found in class color, default attributes used)
(note:  named style gr6 not found in class color, default attributes used)
(note:  named style gr6 not found in class color, default attributes used)
Warning: Because the legend borrowed for the combined graph is not "complete",
automatic resizing of legend markers is disabled and grc1leg2 ignores the option -lmsize-.
See the discussion of "known issues" in grc1leg2's help file here.

.         graph export FigureE1.pdf
file FigureE1.pdf saved as PDF format

. * labeling - regression (Table E2)
. eststo clear

. qui eststo: reg is_terror i.tr_frame attention sex agegr i.relid edu oleh i.region if tr_nonviolence==1

. qui eststo: reg is_illegitimate i.tr_frame attention sex agegr i.relid edu oleh i.region if tr_nonviolence==1

. qui eststo: reg use_force i.tr_frame attention sex agegr i.relid edu oleh i.region if tr_nonviolence==1

. esttab using TableE2.rtf, ///
>         b(3) se(3) r2(3) nogaps compress ///
>         drop(0* 1.relid 1.region) ///
>         star(+ 0.1 * 0.05 ** 0.01 *** 0.001) ///
>         order(_cons 1.tr_frame attention) ///
>         coeflabels(1.tr_frame Terror-Label attention Attention _cons Constant-No_Label ///
>                 sex Gender-Female agegr Age-Group ///
>                 2.relid Religious-Id-Traditional 3.relid Religious-Id-Religious 4.relid ///
>                 Religious-Id-Haredi edu Education oleh Immigrant 2.region Region-North 3.region Region-Haifa ///
>                 4.region Region-Center 5.region Region-Tel-Aviv 6.region Region-South 7.region Region-West-Bank) ///
>         mtitles("Terror Denotation" "Illegitimacy" "Use Force")
(output written to TableE2.rtf)

. * labeling - hetereognous effects, predicted value graphs (Figure E2)
. qui reg is_terror i.tr_frame##i.vote_bloc_rcl i.attention sex agegr i.relid edu oleh i.region if vote_bloc_rcl!=4 & tr_nonviolence==1

. margins, at(vote_bloc_rcl=(1(1)3) tr_frame=(0 1))

Predictive margins                                       Number of obs = 1,178
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_frame      = 0
       vote_bloc_rcl = 1
2._at: tr_frame      = 0
       vote_bloc_rcl = 2
3._at: tr_frame      = 0
       vote_bloc_rcl = 3
4._at: tr_frame      = 1
       vote_bloc_rcl = 1
5._at: tr_frame      = 1
       vote_bloc_rcl = 2
6._at: tr_frame      = 1
       vote_bloc_rcl = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |   4.418806   .3024931    14.61   0.000      3.82531    5.012302
          2  |   6.287737   .2151338    29.23   0.000     5.865642    6.709833
          3  |   7.943381   .1337373    59.40   0.000     7.680986    8.205775
          4  |   4.745961   .3083808    15.39   0.000     4.140913    5.351009
          5  |   7.088509   .2334591    30.36   0.000     6.630458    7.546559
          6  |   8.251725   .1283849    64.27   0.000     7.999832    8.503618
------------------------------------------------------------------------------

. mplotoffset, recast(scatter) ///
>         plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>         plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>         legend(order(3 "No Label" 4 "Terror Label" )) ///
>         yscale(range(5 10)) ylabel(5(2)10, labsize(medsmall)) ///
>         xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(5.5, lpattern(dash) lcolor(gr6)) ///
>         title("Denotation", size(large)) ///
>         name(terror_label_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_frame
(note:  named style gr6 not found in class color, default attributes used)

. qui reg is_illegitimate i.tr_frame##i.vote_bloc_rcl i.attention sex agegr i.relid edu oleh i.region if vote_bloc_rcl!=4 & tr_nonviolence==1

. margins, at(vote_bloc_rcl=(1(1)3) tr_frame=(0 1))

Predictive margins                                       Number of obs = 1,178
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_frame      = 0
       vote_bloc_rcl = 1
2._at: tr_frame      = 0
       vote_bloc_rcl = 2
3._at: tr_frame      = 0
       vote_bloc_rcl = 3
4._at: tr_frame      = 1
       vote_bloc_rcl = 1
5._at: tr_frame      = 1
       vote_bloc_rcl = 2
6._at: tr_frame      = 1
       vote_bloc_rcl = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |   5.606283   .3071296    18.25   0.000      5.00369    6.208876
          2  |   7.315978   .2184314    33.49   0.000     6.887413    7.744544
          3  |   8.203172   .1357872    60.41   0.000     7.936755    8.469589
          4  |     5.7511   .3131076    18.37   0.000     5.136778    6.365422
          5  |   7.474714   .2370375    31.53   0.000     7.009642    7.939785
          6  |   8.293235   .1303527    63.62   0.000     8.037481    8.548989
------------------------------------------------------------------------------

. mplotoffset, recast(scatter) ///
>         plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>         plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>         legend(order(3 "No Label" 4 "Terror Label" )) ///
>         yscale(range(5 10)) ylabel(5(2)10, labsize(medsmall)) ///
>         xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(5.5, lpattern(dash) lcolor(gr6)) ///
>         title("Illegitimacy", size(large)) ///
>         name(legitimate_label_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_frame
(note:  named style gr6 not found in class color, default attributes used)

. qui reg use_force i.tr_frame##i.vote_bloc_rcl i.attention sex agegr i.relid edu oleh i.region if vote_bloc_rcl!=4 & tr_nonviolence==1

. margins, at(vote_bloc_rcl=(1(1)3) tr_frame=(0 1))

Predictive margins                                       Number of obs = 1,178
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_frame      = 0
       vote_bloc_rcl = 1
2._at: tr_frame      = 0
       vote_bloc_rcl = 2
3._at: tr_frame      = 0
       vote_bloc_rcl = 3
4._at: tr_frame      = 1
       vote_bloc_rcl = 1
5._at: tr_frame      = 1
       vote_bloc_rcl = 2
6._at: tr_frame      = 1
       vote_bloc_rcl = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |   4.826729   .2962731    16.29   0.000     4.245437    5.408021
          2  |   6.547406   .2107102    31.07   0.000     6.133989    6.960822
          3  |   8.148802   .1309874    62.21   0.000     7.891802    8.405801
          4  |   4.856674   .3020398    16.08   0.000     4.264068    5.449281
          5  |   7.444473   .2286586    32.56   0.000     6.995841    7.893104
          6  |   8.221996    .125745    65.39   0.000     7.975282    8.468709
------------------------------------------------------------------------------

. mplotoffset, recast(scatter) ///
>         plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>         plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>         legend(order(3 "No Label" 4 "Terror Label" )) ///
>         yscale(range(5 10)) ylabel(5(2)10, labsize(medsmall)) ///
>         xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(5.5, lpattern(dash) lcolor(gr6)) ///
>         title("Use of Force", size(large)) ///
>         name(force_label_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_frame
(note:  named style gr6 not found in class color, default attributes used)

. grc1leg2 terror_label_pv legitimate_label_pv force_label_pv ///
>         , rows(1) xcommon ycommon labsize(small) ///
>         b2title("Vote by Partisan Bloc", size(medsmall)) ///
>         l1title("Predicted Values", size(medsmall)) ring(2) ///
>         xsize(6) ysize(3) scale(1.4) ///
>         name(combine_label, replace)
-grc1leg2- working...
(note:  named style gr6 not found in class color, default attributes used)
(note:  named style gr6 not found in class color, default attributes used)
(note:  named style gr6 not found in class color, default attributes used)
Warning: Because the legend borrowed for the combined graph is not "complete",
automatic resizing of legend markers is disabled and grc1leg2 ignores the option -lmsize-.
See the discussion of "known issues" in grc1leg2's help file here.

. graph export FigureE2.pdf
file FigureE2.pdf saved as PDF format

.         
. * controlling for perceived speaker partisanship
. ************************************************
. * action and harm - regression (Table E3)
. eststo clear

. qui eststo: reg is_terror i.tr_nonviolence attention speaker_ideology  

. qui eststo: reg is_terror i.tr_action attention speaker_ideology 

. qui eststo: reg is_illegitimate i.tr_nonviolence attention speaker_ideology 

. qui eststo: reg is_illegitimate i.tr_action attention speaker_ideology 

. qui eststo: reg use_force i.tr_nonviolence attention speaker_ideology 

. qui eststo: reg use_force i.tr_action attention speaker_ideology 

. esttab using TableE3.rtf, ///
>         b(3) se(3) r2(3) nogaps compress ///
>         drop(1.tr_action 0*) ///
>         star(+ 0.1 * 0.05 ** 0.01 *** 0.001) ///
>         order(_cons 1.tr_nonviolence 2.tr_action 3.tr_action 4.tr_action attention) ///
>         coeflabels(1.tr_nonviolence Nonviolence ///
>                 2.tr_action Economic_action 3.tr_action Legal_action 4.tr_action construct_action /// 
>                 attention Attention _cons Constant-Violence speaker_ideology Speaker-Ideology) ///
>         mgroups("Terror Denotation" "Illegitimacy" "Use Force", pattern(1 0 1 0 1 0)) ///
>         mtitles("Violence" "Action" "Violence" "Action" "Violence" "Action") 
(output written to TableE3.rtf)

. * action and harm - hetereognous effects, predicted value graphs (Figure E3)
. qui reg is_terror i.tr_nonviolence##i.vote_bloc_rcl i.attention speaker_ideology if vote_bloc_rcl!=4

. margins, at(vote_bloc_rcl=(1(1)3) tr_nonviolence=(0 1))

Predictive margins                                       Number of obs = 1,278
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_nonviolence = 0
       vote_bloc_rcl  = 1
2._at: tr_nonviolence = 0
       vote_bloc_rcl  = 2
3._at: tr_nonviolence = 0
       vote_bloc_rcl  = 3
4._at: tr_nonviolence = 1
       vote_bloc_rcl  = 1
5._at: tr_nonviolence = 1
       vote_bloc_rcl  = 2
6._at: tr_nonviolence = 1
       vote_bloc_rcl  = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |   8.730054   .3593729    24.29   0.000     8.025024    9.435084
          2  |   9.047712   .2948297    30.69   0.000     8.469305    9.626119
          3  |    9.25367   .1627521    56.86   0.000     8.934377    9.572962
          4  |   4.125975   .2154662    19.15   0.000     3.703266    4.548684
          5  |   6.309119   .1600199    39.43   0.000     5.995187    6.623052
          6  |   8.121929   .0944971    85.95   0.000     7.936541    8.307316
------------------------------------------------------------------------------

. mplotoffset, recast(scatter) ///
>         plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>         plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>         legend(order(3 "Violence" 4 "Non-violence" )) ///
>         yscale(range(2 10)) ylabel(2(2)10, labsize(medsmall)) ///
>         xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(5.5, lpattern(dash) lcolor(gr6)) ///
>         title("Denotation", size(large)) ///
>         name(terror_violence_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_nonviolence
(note:  named style gr6 not found in class color, default attributes used)

. qui reg is_illegitimate i.tr_nonviolence##i.vote_bloc_rcl i.attention speaker_ideology if vote_bloc_rcl!=4

. margins, at(vote_bloc_rcl=(1(1)3) tr_nonviolence=(0 1))

Predictive margins                                       Number of obs = 1,278
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_nonviolence = 0
       vote_bloc_rcl  = 1
2._at: tr_nonviolence = 0
       vote_bloc_rcl  = 2
3._at: tr_nonviolence = 0
       vote_bloc_rcl  = 3
4._at: tr_nonviolence = 1
       vote_bloc_rcl  = 1
5._at: tr_nonviolence = 1
       vote_bloc_rcl  = 2
6._at: tr_nonviolence = 1
       vote_bloc_rcl  = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |   8.540882   .3629314    23.53   0.000     7.828871    9.252893
          2  |   8.823113   .2977491    29.63   0.000     8.238979    9.407248
          3  |   9.309511   .1643636    56.64   0.000     8.987057    9.631965
          4  |   5.350292   .2175997    24.59   0.000     4.923397    5.777186
          5  |   6.998131   .1616044    43.30   0.000      6.68109    7.315172
          6  |   8.329191   .0954328    87.28   0.000     8.141967    8.516414
------------------------------------------------------------------------------

. mplotoffset, recast(scatter) ///
>         plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>         plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>         legend(order(3 "Violence" 4 "Non-violence" )) ///
>         yscale(range(2 10)) ylabel(2(2)10, labsize(medsmall)) ///
>         xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(5.5, lpattern(dash) lcolor(gr6)) ///
>         title("Illegitimacy", size(large)) ///
>         name(legitimate_violence_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_nonviolence
(note:  named style gr6 not found in class color, default attributes used)

. qui reg use_force i.tr_nonviolence##i.vote_bloc_rcl i.attention speaker_ideology if vote_bloc_rcl!=4

. margins, at(vote_bloc_rcl=(1(1)3) tr_nonviolence=(0 1))

Predictive margins                                       Number of obs = 1,278
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_nonviolence = 0
       vote_bloc_rcl  = 1
2._at: tr_nonviolence = 0
       vote_bloc_rcl  = 2
3._at: tr_nonviolence = 0
       vote_bloc_rcl  = 3
4._at: tr_nonviolence = 1
       vote_bloc_rcl  = 1
5._at: tr_nonviolence = 1
       vote_bloc_rcl  = 2
6._at: tr_nonviolence = 1
       vote_bloc_rcl  = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |   8.541422   .3528041    24.21   0.000     7.849279    9.233565
          2  |   8.619403   .2894407    29.78   0.000     8.051569    9.187238
          3  |   9.446077   .1597772    59.12   0.000      9.13262    9.759533
          4  |   4.399402   .2115278    20.80   0.000      3.98442    4.814384
          5  |   6.608251    .157095    42.07   0.000     6.300057    6.916445
          6  |   8.232878   .0927699    88.75   0.000     8.050879    8.414877
------------------------------------------------------------------------------

. mplotoffset, recast(scatter) ///
>         plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>         plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>         legend(order(3 "Violence" 4 "Non-violence" )) ///
>         yscale(range(2 10)) ylabel(2(2)10, labsize(medsmall)) ///
>         xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(5.5, lpattern(dash) lcolor(gr6)) ///
>         title("Use of Force", size(large)) ///
>         name(force_violence_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_nonviolence
(note:  named style gr6 not found in class color, default attributes used)

. grc1leg2 terror_violence_pv legitimate_violence_pv force_violence_pv ///
>         , rows(1) xcommon ycommon labsize(small) ///
>         b2title("Vote by Partisan Bloc", size(medsmall)) ///
>         l1title("Predicted Values",  size(medsmall)) ring(2) ///
>         xsize(6) ysize(3) scale(1.4) ///
>         name(combine_violence, replace)
-grc1leg2- working...
(note:  named style gr6 not found in class color, default attributes used)
(note:  named style gr6 not found in class color, default attributes used)
(note:  named style gr6 not found in class color, default attributes used)
Warning: Because the legend borrowed for the combined graph is not "complete",
automatic resizing of legend markers is disabled and grc1leg2 ignores the option -lmsize-.
See the discussion of "known issues" in grc1leg2's help file here.

. graph export FigureE3.pdf
file FigureE3.pdf saved as PDF format

. * labeling - regression (Table E4)
. eststo clear

. qui eststo: reg is_terror i.tr_frame attention speaker_ideology if tr_nonviolence==1

. qui eststo: reg is_illegitimate i.tr_frame attention speaker_ideology if tr_nonviolence==1

. qui eststo: reg use_force i.tr_frame attention speaker_ideology if tr_nonviolence==1

. esttab using TableE4.rtf, ///
>         b(3) se(3) r2(3) nogaps compress ///
>         drop(0*) ///
>         star(+ 0.1 * 0.05 ** 0.01 *** 0.001) ///
>         order(_cons 1.tr_frame attention) ///
>         coeflabels(1.tr_frame Terror-Label attention Attention _cons Constant-No_Label ///
>                 speaker_ideology Speaker-Ideology) ///
>         mtitles("Terror Denotation" "Legitimacy" "Use Force") 
(output written to TableE4.rtf)

. * labeling - hetereognous effects, predicted value graphs (Figure E4)
. qui reg is_terror i.tr_frame##i.vote_bloc_rcl i.attention speaker_ideology if vote_bloc_rcl!=4 & tr_nonviolence==1

. margins, at(vote_bloc_rcl=(1(1)3) tr_frame=(0 1))

Predictive margins                                         Number of obs = 961
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_frame      = 0
       vote_bloc_rcl = 1
2._at: tr_frame      = 0
       vote_bloc_rcl = 2
3._at: tr_frame      = 0
       vote_bloc_rcl = 3
4._at: tr_frame      = 1
       vote_bloc_rcl = 1
5._at: tr_frame      = 1
       vote_bloc_rcl = 2
6._at: tr_frame      = 1
       vote_bloc_rcl = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |    3.97267   .3300392    12.04   0.000     3.324983    4.620358
          2  |   5.879736   .2326699    25.27   0.000     5.423131     6.33634
          3  |   7.949524   .1464247    54.29   0.000     7.662172    8.236876
          4  |   4.347374   .3172797    13.70   0.000     3.724726    4.970021
          5  |   6.818759   .2494933    27.33   0.000      6.32914    7.308379
          6  |    8.26078   .1382428    59.76   0.000     7.989484    8.532075
------------------------------------------------------------------------------

. mplotoffset, recast(scatter) ///
>         plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>         plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>         legend(order(3 "No Label" 4 "Terror Label" )) ///
>         yscale(range(5 10)) ylabel(5(2)10, labsize(medsmall)) ///
>         xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(5.5, lpattern(dash) lcolor(gr6)) ///
>         title("Denotation", size(large)) ///
>         name(terror_label_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_frame
(note:  named style gr6 not found in class color, default attributes used)

. qui reg is_illegitimate i.tr_frame##i.vote_bloc_rcl i.attention speaker_ideology if vote_bloc_rcl!=4 & tr_nonviolence==1

. margins, at(vote_bloc_rcl=(1(1)3) tr_frame=(0 1))

Predictive margins                                         Number of obs = 961
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_frame      = 0
       vote_bloc_rcl = 1
2._at: tr_frame      = 0
       vote_bloc_rcl = 2
3._at: tr_frame      = 0
       vote_bloc_rcl = 3
4._at: tr_frame      = 1
       vote_bloc_rcl = 1
5._at: tr_frame      = 1
       vote_bloc_rcl = 2
6._at: tr_frame      = 1
       vote_bloc_rcl = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |   5.234389   .3325703    15.74   0.000     4.581734    5.887043
          2  |   6.862895   .2344542    29.27   0.000     6.402789    7.323001
          3  |    8.32917   .1475476    56.45   0.000     8.039614    8.618726
          4  |   5.487354   .3197129    17.16   0.000     4.859931    6.114776
          5  |   7.160141   .2514066    28.48   0.000     6.666767    7.653516
          6  |   8.324183   .1393029    59.76   0.000     8.050807    8.597559
------------------------------------------------------------------------------

. mplotoffset, recast(scatter) ///
>         plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>         plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>         legend(order(3 "No Label" 4 "Terror Label" )) ///
>         yscale(range(5 10)) ylabel(5(2)10, labsize(medsmall)) ///
>         xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(5.5, lpattern(dash) lcolor(gr6)) ///
>         title("Illegitimacy", size(large)) ///
>         name(legitimate_label_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_frame
(note:  named style gr6 not found in class color, default attributes used)

. qui reg use_force i.tr_frame##i.vote_bloc_rcl i.attention speaker_ideology if vote_bloc_rcl!=4 & tr_nonviolence==1

. margins, at(vote_bloc_rcl=(1(1)3) tr_frame=(0 1))

Predictive margins                                         Number of obs = 961
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_frame      = 0
       vote_bloc_rcl = 1
2._at: tr_frame      = 0
       vote_bloc_rcl = 2
3._at: tr_frame      = 0
       vote_bloc_rcl = 3
4._at: tr_frame      = 1
       vote_bloc_rcl = 1
5._at: tr_frame      = 1
       vote_bloc_rcl = 2
6._at: tr_frame      = 1
       vote_bloc_rcl = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |     4.3657   .3242453    13.46   0.000     3.729382    5.002017
          2  |   6.163911   .2285853    26.97   0.000     5.715322      6.6125
          3  |   8.180511   .1438542    56.87   0.000     7.898203    8.462818
          4  |   4.499322   .3117097    14.43   0.000     3.887606    5.111039
          5  |   7.131243   .2451133    29.09   0.000     6.650219    7.612267
          6  |   8.264416   .1358159    60.85   0.000     7.997883    8.530948
------------------------------------------------------------------------------

. mplotoffset, recast(scatter) ///
>         plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>         plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>         legend(order(3 "No Label" 4 "Terror Label" )) ///
>         yscale(range(5 10)) ylabel(5(2)10, labsize(medsmall)) ///
>         xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(5.5, lpattern(dash) lcolor(gr6)) ///
>         title("Use of Force", size(large)) ///
>         name(force_label_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_frame
(note:  named style gr6 not found in class color, default attributes used)

. grc1leg2 terror_label_pv legitimate_label_pv force_label_pv ///
>         , rows(1) xcommon ycommon labsize(small) ///
>         b2title("Vote by Partisan Bloc", size(medsmall)) ///
>         l1title("Predicted Values",  size(medsmall)) ring(2) ///
>         xsize(6) ysize(3) scale(1.4) ///
>         name(combine_label, replace)
-grc1leg2- working...
(note:  named style gr6 not found in class color, default attributes used)
(note:  named style gr6 not found in class color, default attributes used)
(note:  named style gr6 not found in class color, default attributes used)
Warning: Because the legend borrowed for the combined graph is not "complete",
automatic resizing of legend markers is disabled and grc1leg2 ignores the option -lmsize-.
See the discussion of "known issues" in grc1leg2's help file here.

. graph export FigureE4.pdf
file FigureE4.pdf saved as PDF format

. 
. * heterogeneous effects with longer l-r scale (squared)
. *******************************************************
. * action and harm - hetereognous effects, predicted value graphs (Figure E5)
. qui reg is_terror i.tr_nonviolence##c.rl_id##c.rl_id i.attention 

. margins, dydx(tr_nonviolence) at(rl_id=(1(1)7)) 

Average marginal effects                                 Number of obs = 2,005
Model VCE: OLS

Expression: Linear prediction, predict()
dy/dx wrt:  1.tr_nonviolence
1._at: rl_id = 1
2._at: rl_id = 2
3._at: rl_id = 3
4._at: rl_id = 4
5._at: rl_id = 5
6._at: rl_id = 6
7._at: rl_id = 7

-----------------------------------------------------------------------------------
                  |            Delta-method
                  |      dy/dx   std. err.      t    P>|t|     [95% conf. interval]
------------------+----------------------------------------------------------------
0.tr_nonviolence  |  (base outcome)
------------------+----------------------------------------------------------------
1.tr_nonviolence  |
              _at |
               1  |  -.0785856   .2961661    -0.27   0.791    -.6594123    .5022411
               2  |   -.940845   .1510559    -6.23   0.000    -1.237089   -.6446013
               3  |  -1.766597   .1505695   -11.73   0.000    -2.061887   -1.471307
               4  |  -2.555842    .162839   -15.70   0.000    -2.875194    -2.23649
               5  |   -3.30858   .1987319   -16.65   0.000    -3.698323   -2.918836
               6  |   -4.02481   .3642021   -11.05   0.000    -4.739066   -3.310554
               7  |  -4.704533    .665507    -7.07   0.000    -6.009693   -3.399373
-----------------------------------------------------------------------------------
Note: dy/dx for factor levels is the discrete change from the base level.

. marginsplot, ///
>         xlabel(, format(%9.0g) labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(0, lcolor(gs10)) ///
>         plotopts(mcolor(black) lcolor(black)) ciopts(lcolor(black)) ///
>         title("Denotation", size(large)) ///
>         name(terror_violence_pv, replace)

Variables that uniquely identify margins: rl_id

. qui reg is_illegitimate i.tr_nonviolence##c.rl_id##c.rl_id i.attention 

. margins, dydx(tr_nonviolence) at(rl_id=(1(1)7)) 

Average marginal effects                                 Number of obs = 2,005
Model VCE: OLS

Expression: Linear prediction, predict()
dy/dx wrt:  1.tr_nonviolence
1._at: rl_id = 1
2._at: rl_id = 2
3._at: rl_id = 3
4._at: rl_id = 4
5._at: rl_id = 5
6._at: rl_id = 6
7._at: rl_id = 7

-----------------------------------------------------------------------------------
                  |            Delta-method
                  |      dy/dx   std. err.      t    P>|t|     [95% conf. interval]
------------------+----------------------------------------------------------------
0.tr_nonviolence  |  (base outcome)
------------------+----------------------------------------------------------------
1.tr_nonviolence  |
              _at |
               1  |  -.3123673   .3052201    -1.02   0.306    -.9109503    .2862157
               2  |  -.6424375   .1556738    -4.13   0.000    -.9477375   -.3371374
               3  |  -1.095173   .1551725    -7.06   0.000     -1.39949   -.7908558
               4  |  -1.670573   .1678171    -9.95   0.000    -1.999688   -1.341458
               5  |  -2.368638   .2048073   -11.57   0.000    -2.770297    -1.96698
               6  |  -3.189369    .375336    -8.50   0.000     -3.92546   -2.453278
               7  |  -4.132764    .685852    -6.03   0.000    -5.477824   -2.787704
-----------------------------------------------------------------------------------
Note: dy/dx for factor levels is the discrete change from the base level.

. marginsplot, ///
>         xlabel(, format(%9.0g) labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(0, lcolor(gs10)) ///
>         plotopts(mcolor(black) lcolor(black)) ciopts(lcolor(black)) ///
>         title("Illegitimacy", size(large)) ///
>         name(legitimate_violence_pv, replace)

Variables that uniquely identify margins: rl_id

. qui reg use_force i.tr_nonviolence##c.rl_id##c.rl_id i.attention 

. margins, dydx(tr_nonviolence) at(rl_id=(1(1)7)) 

Average marginal effects                                 Number of obs = 2,005
Model VCE: OLS

Expression: Linear prediction, predict()
dy/dx wrt:  1.tr_nonviolence
1._at: rl_id = 1
2._at: rl_id = 2
3._at: rl_id = 3
4._at: rl_id = 4
5._at: rl_id = 5
6._at: rl_id = 6
7._at: rl_id = 7

-----------------------------------------------------------------------------------
                  |            Delta-method
                  |      dy/dx   std. err.      t    P>|t|     [95% conf. interval]
------------------+----------------------------------------------------------------
0.tr_nonviolence  |  (base outcome)
------------------+----------------------------------------------------------------
1.tr_nonviolence  |
              _at |
               1  |  -.2873444   .2908481    -0.99   0.323    -.8577418     .283053
               2  |   -.925933   .1483436    -6.24   0.000    -1.216857   -.6350087
               3  |  -1.565936   .1478659   -10.59   0.000    -1.855923   -1.275948
               4  |  -2.207352    .159915   -13.80   0.000     -2.52097   -1.893735
               5  |  -2.850183   .1951635   -14.60   0.000    -3.232928   -2.467438
               6  |  -3.494428   .3576625    -9.77   0.000    -4.195858   -2.792997
               7  |  -4.140086   .6535572    -6.33   0.000    -5.421811   -2.858361
-----------------------------------------------------------------------------------
Note: dy/dx for factor levels is the discrete change from the base level.

. marginsplot, ///
>         xlabel(, format(%9.0g) labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(0, lcolor(gs10)) ///
>         plotopts(mcolor(black) lcolor(black)) ciopts(lcolor(black)) ///
>         title("Use of Force", size(large)) ///
>         name(force_violence_pv, replace)

Variables that uniquely identify margins: rl_id

. grc1leg2 terror_violence_pv legitimate_violence_pv force_violence_pv ///
>         , rows(1) xcommon ycommon labsize(small) ///
>         b2title("Ideological Self-identification (Right to Left)", size(medsmall)) ///
>         l1title("Marginal Effect of Nonviolent Action",  size(medsmall)) ring(2) ///
>         xsize(6) ysize(3) scale(1.4) ///
>         name(combine_violence, replace) loff
-grc1leg2- working...

. graph export FigureE5.pdf
file FigureE5.pdf saved as PDF format

. * labeling - hetereognous effects, predicted value graphs (Figure E6)
. qui reg is_terror i.tr_frame##c.rl_id##c.rl_id i.attention if tr_nonviolence==1

. margins, dydx(tr_frame) at(rl_id=(1(1)7)) 

Average marginal effects                                 Number of obs = 1,492
Model VCE: OLS

Expression: Linear prediction, predict()
dy/dx wrt:  1.tr_frame
1._at: rl_id = 1
2._at: rl_id = 2
3._at: rl_id = 3
4._at: rl_id = 4
5._at: rl_id = 5
6._at: rl_id = 6
7._at: rl_id = 7

------------------------------------------------------------------------------
             |            Delta-method
             |      dy/dx   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
0.tr_frame   |  (base outcome)
-------------+----------------------------------------------------------------
1.tr_frame   |
         _at |
          1  |  -.3140421   .3069271    -1.02   0.306     -.916099    .2880147
          2  |   .2678327   .1604506     1.67   0.095    -.0469012    .5825666
          3  |   .5660743   .1644133     3.44   0.001     .2435672    .8885814
          4  |   .5806828   .1771287     3.28   0.001     .2332336    .9281319
          5  |   .3116581   .2081178     1.50   0.134     -.096578    .7198941
          6  |  -.2409998   .3723155    -0.65   0.518    -.9713201    .4893205
          7  |  -1.077291   .6813926    -1.58   0.114    -2.413885    .2593034
------------------------------------------------------------------------------
Note: dy/dx for factor levels is the discrete change from the base level.

. marginsplot, ///
>         xlabel(, format(%9.0g) labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(0, lcolor(gs10)) ///
>         plotopts(mcolor(black) lcolor(black)) ciopts(lcolor(black)) ///
>         title("Denotation", size(large)) ///
>         name(terror_label_pv, replace)

Variables that uniquely identify margins: rl_id

. qui reg is_illegitimate i.tr_frame##c.rl_id##c.rl_id i.attention if tr_nonviolence==1

. margins, dydx(tr_frame) at(rl_id=(1(1)7)) 

Average marginal effects                                 Number of obs = 1,492
Model VCE: OLS

Expression: Linear prediction, predict()
dy/dx wrt:  1.tr_frame
1._at: rl_id = 1
2._at: rl_id = 2
3._at: rl_id = 3
4._at: rl_id = 4
5._at: rl_id = 5
6._at: rl_id = 6
7._at: rl_id = 7

------------------------------------------------------------------------------
             |            Delta-method
             |      dy/dx   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
0.tr_frame   |  (base outcome)
-------------+----------------------------------------------------------------
1.tr_frame   |
         _at |
          1  |  -.3024271   .3122618    -0.97   0.333    -.9149483     .310094
          2  |  -.0106701   .1632394    -0.07   0.948    -.3308744    .3095342
          3  |    .154605    .167271     0.92   0.355    -.1735075    .4827175
          4  |   .1933981   .1802074     1.07   0.283      -.16009    .5468862
          5  |   .1057092    .211735     0.50   0.618    -.3096223    .5210408
          6  |  -.1084616   .3787867    -0.29   0.775    -.8514755    .6345523
          7  |  -.4491144   .6932358    -0.65   0.517     -1.80894    .9107111
------------------------------------------------------------------------------
Note: dy/dx for factor levels is the discrete change from the base level.

. marginsplot, ///
>         xlabel(, format(%9.0g) labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(0, lcolor(gs10)) ///
>         plotopts(mcolor(black) lcolor(black)) ciopts(lcolor(black)) ///
>         title("Illegitimacy", size(large)) ///
>         name(legitimate_label_pv, replace)

Variables that uniquely identify margins: rl_id

. qui reg use_force i.tr_frame##c.rl_id##c.rl_id i.attention if tr_nonviolence==1

. margins, dydx(tr_frame) at(rl_id=(1(1)7)) 

Average marginal effects                                 Number of obs = 1,492
Model VCE: OLS

Expression: Linear prediction, predict()
dy/dx wrt:  1.tr_frame
1._at: rl_id = 1
2._at: rl_id = 2
3._at: rl_id = 3
4._at: rl_id = 4
5._at: rl_id = 5
6._at: rl_id = 6
7._at: rl_id = 7

------------------------------------------------------------------------------
             |            Delta-method
             |      dy/dx   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
0.tr_frame   |  (base outcome)
-------------+----------------------------------------------------------------
1.tr_frame   |
         _at |
          1  |  -.1202964   .3037325    -0.40   0.692    -.7160868     .475494
          2  |   .1660313   .1587806     1.05   0.296    -.1454267    .4774894
          3  |   .3101789   .1627021     1.91   0.057    -.0089714    .6293292
          4  |   .3121464   .1752851     1.78   0.075    -.0316864    .6559792
          5  |   .1719337   .2059516     0.83   0.404    -.2320532    .5759207
          6  |   -.110459   .3684403    -0.30   0.764    -.8331779    .6122598
          7  |   -.535032   .6743004    -0.79   0.428    -1.857714    .7876505
------------------------------------------------------------------------------
Note: dy/dx for factor levels is the discrete change from the base level.

. marginsplot, ///
>         xlabel(, format(%9.0g) labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(0, lcolor(gs10)) ///
>         plotopts(mcolor(black) lcolor(black)) ciopts(lcolor(black)) ///
>         title("Use of Force", size(large)) ///
>         name(force_label_pv, replace)

Variables that uniquely identify margins: rl_id

. grc1leg2 terror_label_pv legitimate_label_pv force_label_pv ///
>         , rows(1) xcommon ycommon labsize(small) ///
>         b2title("Ideological Self-identification (Right to Left)", size(medsmall)) ///
>         l1title("Marginal Effect of Terror Label",  size(medsmall)) ring(2) ///
>         xsize(6) ysize(3) scale(1.4) ///
>         name(combine_label, replace) loff
-grc1leg2- working...

. graph export FigureE6.pdf
file FigureE6.pdf saved as PDF format

. 
. * collapsed DV scales
. *********************
. * generate collapsed scales:
. foreach var of varlist is_terror is_illegitimate use_force {
  2.         recode `var' (1/2 = 1)(3/4 = 2)(5/6 = 3)(7/8 = 4)(9/10 = 5), gen(`var'_5)
  3. }
(1,911 differences between is_terror and is_terror_5)
(1,934 differences between is_illegitimate and is_illegitimate_5)
(1,932 differences between use_force and use_force_5)

. * action and harm - regression (Table E5)
. eststo clear

. qui eststo: reg is_terror_5 i.tr_nonviolence attention

. qui eststo: reg is_terror_5 i.tr_action attention

. qui eststo: reg is_illegitimate_5 i.tr_nonviolence attention

. qui eststo: reg is_illegitimate_5 i.tr_action attention

. qui eststo: reg use_force_5 i.tr_nonviolence attention

. qui eststo: reg use_force_5 i.tr_action attention

. esttab using TableE5.rtf, ///
>         b(3) se(3) r2(3) nogaps compress ///
>         drop(1.tr_action 0*) ///
>         star(+ 0.1 * 0.05 ** 0.01 *** 0.001) ///
>         order(_cons 1.tr_nonviolence 2.tr_action 3.tr_action 4.tr_action attention) ///
>         coeflabels(1.tr_nonviolence Nonviolence ///
>                 2.tr_action Economic_action 3.tr_action Legal_action 4.tr_action Construct_action /// 
>                 1.attention Attention _cons Constant-Violence) ///
>         mgroups("Terror Denotation" "Illegitimacy" "Use Force" pattern(1 0 1 0 1 0)) ///
>         mtitles("Violence" "Action" "Violence" "Action" "Violence" "Action")
(output written to TableE5.rtf)

. * action and harm - hetereognous effects, predicted value graphs (Figure E7)
. qui reg is_terror_5 i.tr_nonviolence##i.vote_bloc_rcl i.attention if vote_bloc_rcl!=4

. margins, at(vote_bloc_rcl=(1(1)3) tr_nonviolence=(0 1))

Predictive margins                                       Number of obs = 1,583
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_nonviolence = 0
       vote_bloc_rcl  = 1
2._at: tr_nonviolence = 0
       vote_bloc_rcl  = 2
3._at: tr_nonviolence = 0
       vote_bloc_rcl  = 3
4._at: tr_nonviolence = 1
       vote_bloc_rcl  = 1
5._at: tr_nonviolence = 1
       vote_bloc_rcl  = 2
6._at: tr_nonviolence = 1
       vote_bloc_rcl  = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |   4.563969   .1504424    30.34   0.000     4.268881    4.859057
          2  |   4.655834   .1157184    40.23   0.000     4.428856    4.882813
          3  |   4.685826   .0666422    70.31   0.000     4.555109    4.816543
          4  |   2.436571   .0910564    26.76   0.000     2.257966    2.615175
          5  |   3.500329   .0666386    52.53   0.000     3.369619    3.631039
          6  |   4.269308   .0390722   109.27   0.000     4.192669    4.345947
------------------------------------------------------------------------------

. mplotoffset, recast(scatter) ///
>         plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>         plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>         legend(order(3 "Violence" 4 "Non-violence" )) ///
>         yscale(range(1 5)) ylabel(1(1)5, labsize(medsmall)) ///
>         xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(3, lpattern(dash) lcolor(gr6)) ///
>         title("Denotation", size(large)) ///
>         name(terror_violence_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_nonviolence
(note:  named style gr6 not found in class color, default attributes used)

. qui reg is_illegitimate_5 i.tr_nonviolence##i.vote_bloc_rcl i.attention if vote_bloc_rcl!=4

. margins, at(vote_bloc_rcl=(1(1)3) tr_nonviolence=(0 1))

Predictive margins                                       Number of obs = 1,583
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_nonviolence = 0
       vote_bloc_rcl  = 1
2._at: tr_nonviolence = 0
       vote_bloc_rcl  = 2
3._at: tr_nonviolence = 0
       vote_bloc_rcl  = 3
4._at: tr_nonviolence = 1
       vote_bloc_rcl  = 1
5._at: tr_nonviolence = 1
       vote_bloc_rcl  = 2
6._at: tr_nonviolence = 1
       vote_bloc_rcl  = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |   4.525058     .15374    29.43   0.000     4.223502    4.826615
          2  |   4.543244   .1182549    38.42   0.000      4.31129    4.775198
          3  |   4.674304    .068103    68.64   0.000     4.540722    4.807886
          4  |   2.972289   .0930523    31.94   0.000      2.78977    3.154809
          5  |   3.757044   .0680993    55.17   0.000     3.623469    3.890619
          6  |   4.328818   .0399286   108.41   0.000       4.2505    4.407137
------------------------------------------------------------------------------

. mplotoffset, recast(scatter) ///
>         plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>         plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>         legend(order(3 "Violence" 4 "Non-violence" )) ///
>         yscale(range(1 5)) ylabel(1(1)5, labsize(medsmall)) ///
>         xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(3, lpattern(dash) lcolor(gr6)) ///
>         title("Illegitimacy", size(large)) ///
>         name(legitimate_violence_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_nonviolence
(note:  named style gr6 not found in class color, default attributes used)

. qui reg use_force_5 i.tr_nonviolence##i.vote_bloc_rcl i.attention if vote_bloc_rcl!=4

. margins, at(vote_bloc_rcl=(1(1)3) tr_nonviolence=(0 1))

Predictive margins                                       Number of obs = 1,583
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_nonviolence = 0
       vote_bloc_rcl  = 1
2._at: tr_nonviolence = 0
       vote_bloc_rcl  = 2
3._at: tr_nonviolence = 0
       vote_bloc_rcl  = 3
4._at: tr_nonviolence = 1
       vote_bloc_rcl  = 1
5._at: tr_nonviolence = 1
       vote_bloc_rcl  = 2
6._at: tr_nonviolence = 1
       vote_bloc_rcl  = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |   4.429532   .1482226    29.88   0.000     4.138798    4.720266
          2  |   4.519055    .114011    39.64   0.000     4.295426    4.742684
          3  |   4.787793   .0656589    72.92   0.000     4.659005    4.916581
          4  |   2.534961   .0897128    28.26   0.000     2.358992     2.71093
          5  |   3.605945   .0656554    54.92   0.000     3.477164    3.734726
          6  |   4.304387   .0384957   111.81   0.000     4.228879    4.379895
------------------------------------------------------------------------------

. mplotoffset, recast(scatter) ///
>         plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>         plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>         legend(order(3 "Violence" 4 "Non-violence" )) ///
>         yscale(range(1 5)) ylabel(1(1)5, labsize(medsmall)) ///
>         xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(3, lpattern(dash) lcolor(gr6)) ///
>         title("Use of Force", size(large)) ///
>         name(force_violence_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_nonviolence
(note:  named style gr6 not found in class color, default attributes used)

. grc1leg2 terror_violence_pv legitimate_violence_pv force_violence_pv ///
>         , rows(1) xcommon ycommon labsize(small) ///
>         b2title("Vote by Partisan Bloc", size(medsmall)) ///
>         l1title("Predicted Values",  size(medsmall)) ring(2) ///
>         xsize(6) ysize(3) scale(1.4) ///
>         name(combine_violence_5, replace)
-grc1leg2- working...
(note:  named style gr6 not found in class color, default attributes used)
(note:  named style gr6 not found in class color, default attributes used)
(note:  named style gr6 not found in class color, default attributes used)
Warning: Because the legend borrowed for the combined graph is not "complete",
automatic resizing of legend markers is disabled and grc1leg2 ignores the option -lmsize-.
See the discussion of "known issues" in grc1leg2's help file here.

. graph export FigureE7.pdf
file FigureE7.pdf saved as PDF format

. * labeling - regression (Table E6)
. eststo clear

. qui eststo: reg is_terror_5 i.tr_frame attention if tr_nonviolence==1

. qui eststo: reg is_illegitimate_5 i.tr_frame attention if tr_nonviolence==1

. qui eststo: reg use_force_5 i.tr_frame attention if tr_nonviolence==1

. esttab using TableE6.rtf, ///
>         b(3) se(3) r2(3) nogaps compress ///
>         drop(0*) ///
>         star(+ 0.1 * 0.05 ** 0.01 *** 0.001) ///
>         order(_cons 1.tr_frame attention) ///
>         coeflabels(1.tr_frame Terror-Label attention Attention _cons Constant-No_Label) ///
>         mtitles("Terror Denotation" "Legitimacy" "Use Force")
(output written to TableE6.rtf)

. * labeling - hetereognous effects, predicted value graphs (Figure E8)
. qui reg is_terror_5 i.tr_frame##i.vote_bloc_rcl i.attention if vote_bloc_rcl!=4 & tr_nonviolence==1

. margins, at(vote_bloc_rcl=(1(1)3) tr_frame=(0 1))

Predictive margins                                       Number of obs = 1,178
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_frame      = 0
       vote_bloc_rcl = 1
2._at: tr_frame      = 0
       vote_bloc_rcl = 2
3._at: tr_frame      = 0
       vote_bloc_rcl = 3
4._at: tr_frame      = 1
       vote_bloc_rcl = 1
5._at: tr_frame      = 1
       vote_bloc_rcl = 2
6._at: tr_frame      = 1
       vote_bloc_rcl = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |   2.393607   .1365618    17.53   0.000     2.125674     2.66154
          2  |   3.320673   .0967883    34.31   0.000     3.130775    3.510571
          3  |   4.205545   .0600906    69.99   0.000     4.087647    4.323442
          4  |   2.487716   .1383418    17.98   0.000      2.21629    2.759141
          5  |   3.712388   .1047862    35.43   0.000     3.506798    3.917978
          6  |   4.326239   .0578763    74.75   0.000     4.212686    4.439792
------------------------------------------------------------------------------

. mplotoffset, recast(scatter) ///
>         plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>         plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>         legend(order(3 "No Label" 4 "Terror Label" )) ///
>         yscale(range(1 5)) ylabel(1(1)5, labsize(medsmall)) ///
>         xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(3, lpattern(dash) lcolor(gr6)) ///
>         title("Denotation", size(large)) ///
>         name(terror_label_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_frame
(note:  named style gr6 not found in class color, default attributes used)

. qui reg is_illegitimate_5 i.tr_frame##i.vote_bloc_rcl i.attention if vote_bloc_rcl!=4 & tr_nonviolence==1

. margins, at(vote_bloc_rcl=(1(1)3) tr_frame=(0 1))

Predictive margins                                       Number of obs = 1,178
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_frame      = 0
       vote_bloc_rcl = 1
2._at: tr_frame      = 0
       vote_bloc_rcl = 2
3._at: tr_frame      = 0
       vote_bloc_rcl = 3
4._at: tr_frame      = 1
       vote_bloc_rcl = 1
5._at: tr_frame      = 1
       vote_bloc_rcl = 2
6._at: tr_frame      = 1
       vote_bloc_rcl = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |    2.95613   .1392533    21.23   0.000     2.682916    3.229344
          2  |   3.739495   .0986959    37.89   0.000     3.545854    3.933136
          3  |   4.313885    .061275    70.40   0.000     4.193664    4.434106
          4  |   2.989226   .1410684    21.19   0.000     2.712451    3.266001
          5  |   3.777413   .1068514    35.35   0.000     3.567771    3.987055
          6  |   4.342174    .059017    73.57   0.000     4.226383    4.457964
------------------------------------------------------------------------------

. mplotoffset, recast(scatter) ///
>         plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>         plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>         legend(order(3 "No Label" 4 "Terror Label" )) ///
>         yscale(range(1 5)) ylabel(1(1)5, labsize(medsmall)) ///
>         xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(3, lpattern(dash) lcolor(gr6)) ///
>         title("Illegitimacy", size(large)) ///
>         name(legitimate_label_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_frame
(note:  named style gr6 not found in class color, default attributes used)

. qui reg use_force_5 i.tr_frame##i.vote_bloc_rcl i.attention if vote_bloc_rcl!=4 & tr_nonviolence==1

. margins, at(vote_bloc_rcl=(1(1)3) tr_frame=(0 1))

Predictive margins                                       Number of obs = 1,178
Model VCE: OLS

Expression: Linear prediction, predict()
1._at: tr_frame      = 0
       vote_bloc_rcl = 1
2._at: tr_frame      = 0
       vote_bloc_rcl = 2
3._at: tr_frame      = 0
       vote_bloc_rcl = 3
4._at: tr_frame      = 1
       vote_bloc_rcl = 1
5._at: tr_frame      = 1
       vote_bloc_rcl = 2
6._at: tr_frame      = 1
       vote_bloc_rcl = 3

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         _at |
          1  |   2.570954   .1349785    19.05   0.000     2.306128    2.835781
          2  |   3.397245   .0956661    35.51   0.000     3.209549    3.584941
          3  |   4.289345   .0593939    72.22   0.000     4.172815    4.405876
          4  |   2.502702   .1367379    18.30   0.000     2.234423     2.77098
          5  |   3.851478   .1035713    37.19   0.000     3.648272    4.054684
          6  |   4.316672   .0572053    75.46   0.000     4.204436    4.428908
------------------------------------------------------------------------------

. mplotoffset, recast(scatter) ///
>         plot1opts(mcolor(black)) ci1opts(lcolor(black)) ///
>         plot2opts(mcolor(black) msymbol(Oh)) ci2opts(lcolor(black)) ///
>         legend(order(3 "No Label" 4 "Terror Label" )) ///
>         yscale(range(1 5)) ylabel(1(1)5, labsize(medsmall)) ///
>         xlabel(1 "Left" 2 "Center" 3 "Right", labsize(medsmall)) ///
>         ytitle("")      xtitle("")  ///
>         yline(3, lpattern(dash) lcolor(gr6)) ///
>         title("Use of Force", size(large)) ///
>         name(force_label_pv, replace)

  Variables that uniquely identify margins: vote_bloc_rcl tr_frame
(note:  named style gr6 not found in class color, default attributes used)

. grc1leg2 terror_label_pv legitimate_label_pv force_label_pv ///
>         , rows(1) xcommon ycommon labsize(small) ///
>         b2title("Vote by Partisan Bloc", size(medsmall)) ///
>         l1title("Predicted Values",  size(medsmall)) ring(2) ///
>         xsize(6) ysize(3) scale(1.4) ///
>         name(combine_label, replace)
-grc1leg2- working...
(note:  named style gr6 not found in class color, default attributes used)
(note:  named style gr6 not found in class color, default attributes used)
(note:  named style gr6 not found in class color, default attributes used)
Warning: Because the legend borrowed for the combined graph is not "complete",
automatic resizing of legend markers is disabled and grc1leg2 ignores the option -lmsize-.
See the discussion of "known issues" in grc1leg2's help file here.

. graph export FigureE8.pdf
file FigureE8.pdf saved as PDF format

. 
end of do-file

. log close
      name:  <unnamed>
       log:  C:\Users\alony\Dropbox\Alon\@ Research projects @\Peace process opinion\Perceptions of terrorism (with Avishay BSG)\Submissions\2022
> .09 PSRM (accepted)\2023.04 Final version\Data replication files\Replication log - Stata - Israel.log
  log type:  text
 closed on:  16 May 2023, 18:45:55
-------------------------------------------------------------------------------------------------------------------------------------------------
